A software tool for the simultaneous analysis of cortical slow waves in electrophysiological and laser-speckle hemodynamic recordings of the murine lesion model

Abstract

Cortical slow waves (< 2 Hz) can be detected from the firing of cortical neurons, and they also produce a slower change in cerebral blood flow. These two readings are usually done apart, and almost never on the same animal. We present a software tool that handles both at the same time, for the murine focal-lesion model. The tool reads a 32-channel micro-electrode-array (MEA) recording of local field potentials (LFP, Open Ephys, 5 kHz) and wide-field laser speckle contrast images of regional cerebral blood flow, pre-processed by the ICFO Medical Optics group. We describe its four signal-processing blocks: (i) zero-phase Butterworth band-pass filtering of the LFP; (ii) Up/Down state detection on a score that fuses a log-MUA estimate and a high-band variance estimate by principal-component analysis, with an Otsu threshold and a minimum-duration state cleaner; (iii) a wave-propagation block that groups channels by cross-correlation coherence with hierarchical clustering and fits a two-dimensional plane wave over the MEA geometry to get the propagation speed and direction; and (iv) a viewer that re-aligns the channels by the lags from the propagation block. The blocks are run on real LFP recordings of an anesthetized mouse, and all figures are produced by the same code used by the interface. The tool also implements a fifth block that links the electrical and the hemodynamic signals to map neurovascular coupling; it is implemented but not yet tested on real data, since we do not have a recording with both signals from the same animal yet.